and Huai Bong areas, where south to southwest facing slopes are much drier than
those opposite. The same results were found in Bor Krai, but were much less
pronounced.
Petrography explained more than 55 % of the variance in Bor Krai, with
Limestone strongly related to the presence of Acrisols and Leptosols, and
claystone, siltstone and sandstone strongly related to the presence of Alisols.
Almost all the Luvisols were located on latite. In Huai Bong, petrography helped
to distinguish between Cambisols on alluvial deposits, Regosols and Leptosols on
shale, and Alisols on clastic sediments, while in the Mae Sa watershed, it helped to
separate-out Acrisols, Umbrisols, Cambisols and Gleysols. Acrisols clearly
dominated on migmatite and gneiss, while Umbrisols and Cambisols prevailed on
marble. Gleysols prevailed on Tertiary and Quaternary sediments in broader valleys
and in the Chiang Mai basin. Also important were satellite data (see Table 2.3), and
in the Mae Sa watershed SPOT 5 band 2 was essential for distinguishing between
Technosols, water bodies and other soil types. In Huai Bong, this helped to
distinguish between Cambisols with bare surfaces in the valley bottom and other
soil types, which were mostly covered with forest. SPOT 5 band 3 helped to
distinguish between Anthrosols, Umbrisols, Cambisols and water bodies and
other mapping units in the Mae Sa watershed. Anthrosols most probably correlated
with low biomass, while the locations of Umbrisols and Cambisols mostly
corresponded with undisturbed evergreen forest with high biomass. SPOT 5 band
4 was essential in the Mae Sa watershed for discriminating between Umbrisols and
Cambisols under evergreen forest on the one hand, and Acrisols under cultivation
and deciduous dipterocarp forest on the other. In Huai Bong, Cambisols correlated
very well with harvested paddy fields, showing high albedo and evergreen trees
along streams – representing low albedo. Regosols and Leptosols correlated with
deciduous forest due to the limited soil thickness, with rather high reflections,
whereas Alisols dominated on gentle slopes under moderate deciduous dipterocarp
forests with fairly low reflections. In Bor Krai, the vegetation moisture slightly
increased with elevation, corresponding with a decrease in reflection, and with the
transition from Alisols to Acrisols occurring at around 850 m.a.s.l.
In the Mae Sa watershed, the slope inclination helped to separate between
Gleysols, Umbrisols, Cambisols, Leptosols and other soils. While Gleysols were
found mostly on very gentle slopes, Umbrisols, Cambisols, and Leptosols were
found on rather steep slopes. In Huai Bong, slope inclination was very important in
helping to distinguish between Cambisols along the valley bottom, Leptosols and
Regosols on extremely steep slopes and Alisols on moderate slopes. In Bor Krai,
slope inclination mainly separated nudilithic Leptosols (limestone outcrops) and
Leptosols from other map units.
The classification tree approach yielded high levels of accuracy – at about the
same level as when using the maximum likelihood method. One advantage the
classification tree mapping approach has over the maximum likelihood approach is
its greater transparency, making its interpretation easier (McBratney et al. 2003).
Furthermore, it is possible to implement the revealed classification rules and
probabilities within expert systems, such as SoLIM (Zhu et al. 2001), plus
60
K. Stahr et al.
Précédent

- 68/490

Suivant